1．Introduction

　Two major technology trends are reshaping the telecommunication industry today. On one hand, more and more services are moving into the cloud enabled by the ubiquitousness of available communication links. On the other hand, end user devices are becoming increasingly powerful and mobile, enabling unforeseen number of new applications making life easier and saving users the most precious commodity, time. Traditional telecommunication equipments are moving from a hardware solution to the cloud solution in the form of Network Function Virtualization (NVF).

　Telecommunication service providers are uniquely positioned close to the end user to enhance existing cloud services and enable a new breed of services. These services are not readily available today because of their demand for both extremely short latency and extremely high computation power and access to extremely large amount of data.

2．Data Center Alone is Not Enough

　Worldwide shipment of smart mobile devices such as smart phones and tablets today has already surpassed desktop devices recently(1)^,(2). Furthermore, the total number of smart phones, including phablets will be nearly half of global devices by 2020(3).

　The trend of mobile device growth will continue as the devices themselves become increasingly powerful both in computation power as in multi-core CPUs and storage capacity as in multi-GB RAMs. However, the desirable properties of mobile devices remain unchanged and are unlikely to change in the future. They need to be small in size, light in weight and long in battery life. These properties result in inherit limitation on the growth of compute power and storage capacity of mobile devices compared to servers with fixed location and connection to the network(4).

　Today’s mobile devices are valuable to end users because of their access to traditional cloud services such as Facebook, YouTube, Netflix, Google Maps. With the inclusion of sensors such as GPS, video cameras, gyroscope sensors, accelerometers on smart phones, a new breed of applications has the potential to be realized on these devices. One example is perception-based applications that enhance users’ understanding of the physical world through the sensors on their mobile devices such as augmented reality, face recognition, speech to text, etc.

　However, these applications are not readily available today because mobile devices are not powerful enough to process rich media, to store large amount of data or to execute data analysis algorithms required for these applications. The natural solution is to offload computation and data storage to the cloud to enable new mobile applications that are compute intensive and rely on access to large amount of data.